Contact: Bruce Robinson MS F665, Los Alamos National Laboratory Los Alamos, NM 87545 Voice (505)667-1910 FAX (505)665-8737 robinson@lanl.gov

Porous Media Flow Code Development

Many of the world's energy and environmental concerns can be addressed by modeling the flow of fluids and transport of dissolved materials beneath the earth's surface (e.g., oil and gas reservoirs, contaminated aquifers). The performance of these tasks requires the numerical tools to quantitatively model flow and transport through porous/fractured media.

The Porous and Fracture Flow team of EES-5 is active in research and applications in many areas concerning subsurface flow and transport. Projects are characterized by a wide spectrum of time and distance scales, ranging from hydrothermal circulation at mid-ocean ridge spreading centers, contaminant transport at environmental restoration sites, high level radioactive waste disposal, containment of underground nuclear tests, and pore scale processes.

---

Codes and Computational Tools

The Porous and Fracture Flow team develops, modifies, and supports sophisticated numerical simulation codes for subsurface transport processes (FEHM and TRACR3D). These flow and transport codes interface with the GEOMESH and X3D grid generation tools which aid in the construction of complex multi-material finite element and finite difference computational grids. The codes model three-dimensional, time-dependent, multiphase, multicomponent, nonisothermal, reactive flow through porous and fractured media. Capabilities also include the ability to accurately represent complex 3-D geologic media and structures and their effects on subsurface flow and transport.

GEOMESH LaGriT Subsurface Transport Process (FEHM) Permeability Models

---

Table 1:Major Computer Codes in Porous and Fracture Flow Section

FEHM (Finite Element Heat and Mass) 3-D flow of gas ,water, oil, and heat flow of air, water, and heat multiple chemically reactive and sorbing tracers finite element/finite volume formulation coupled stress module saturated and unsaturated media preconditioned conjugate gradient solution of coupled linear equations fully implicit, fully coupled Newton Raphson solution of nonlinear equations double porosity and double porosity/double permeability capabilities

TRACR3D 3-D flow of air and water multiple chemically reactive, radioactive and sorbing tracers integrated finite difference formulation saturated and unsaturated media preconditioned conjugate gradient solution of linear equations fully implicit Newton Raphson solution of nonlinear equations biokinetics adjoint sensitivity capabilities

To meet the needs of such a diverse project base, our section has developed several large scale subsurface flow simulation computer codes over the past decade (Table 1) and is continually involved in developing new techniques to improve computational efficiency and take advantage of evolving computer architectures. For example, we are currently porting our simulation codes to Thinking Machines' new Connection Machine-5, the most powerful parallel computer in the world. In addition to being state-of-the-art in numerical techniques, our codes are also state-of-the-art in physical representation of real-world problems (i.e., the capability to model in situ bioremediation). In the area of flow and transport in fractured rock, we are collaborating with other scientists in the Laboratory and with the Canadian Atomic Energy Commission Whiteshell Laboratory in Pinawa, Canada in a combined experimental/theoretical study of flow in fractured rock. Comparisons are being made between theories of dispersion and tracer tests in machined and natural fractures. In addition, the structure of fractures is being modeled with fractal mathematics and will be compared to profiles measured in real fractures. This project will support development of novel approaches to fracture flow, such as fractal characterization and inversion algorithms, and the use of cellular automata algorithms.

---

Figure 1. Steady state saturation in a site scale model of Yucca Mountain, Nevada. The area of this model covers a 1.7X4.2 square mile area. Variation in saturation very closely mirrors variations in rock properties. Model calculations like this are used for evaluating the suitability of Yucca Mountain as a potential repository for high level nuclear waste.

The Yucca Mountain Project has funded much of the code development and quality control procedures of FEHM and TRACR3D codes used in the section. This project studies the safety of storing high level radioactive waste in an underground repository above the water table in a relatively dry environment. In order to model the retardation potential of the repository rock at Yucca Mountain, a coordinated effort between researchers in geochemistry and groundwater modeling was necessary. Using a database created by Sandia National Laboratory which describes some 14 stratigraphic units at Yucca Mountain, flow and transport properties based on rock type and radionuclide type were assigned to 57,000 nodes in a 3-D grid. Figure 1 shows a color rendering of the steady state saturation field at Yucca Mountain . Note the 3-D nature of the flow field. In fact, the studies revealed fast paths not available in 1-D or 2-D simulations.




---

In the last few years, we have emphasized the area of grid generation and geologic framework modeling. It is increasing necessary to coordinate the often 3-d conceptual model generated by geologists and other field personnel with the computer model run by a reservoir engineer. The conceptual model is often generated and viewed with advanced GIS software and may contain millions of nodes. To run the numerical reservoir model the conceptual model must be coarsened to contain several hundred thousand nodes to be tractable in a workstation environment. The very high resolution conceptual model can usually display realistically faulting, pinchouts and rapidly changing topography. When coarsened with structured (finite difference) grids, information is often lost because of the "blocky" nature of the grids and the averaging that must take place. This limitation is overcome by the use of unstructured (finite element) grids which are able to fit geometry well even with low resolution grids (Figure 2).

Figure 2. Comparison of a two different gridding methods to model the same geometry. The structured grid made up of uniform size bricks must stair step to follow non-horizontal layers. The layer thickness is also not accurately represented. The unstructured tetrahedral grid on the right exactly matches the geometry of the input data.

EES-5 | EES Division | LANL | DOE
Phone Book | Search

L O S   A L A M O S   N A T I O N A L   L A B O R A T O R Y
Operated by the University of California for the US Department of Energy